CN104636371A - Information recommendation method and device - Google Patents

Abstract

The embodiment of the present invention discloses an information recommendation method and device, wherein the method includes: extracting feature factors from the social media of the user and/or the user's friends; The information pointed to by the index information; the feature factor is extracted from the information pointed to by the index information; according to the extracted feature factor in social media and the feature factor in the information pointed to by the index information and the user’s point to the index information The preference score of information obtains the user's preference singular value SVD model; the characteristic factors in the social media and the characteristic factors of the information to be recommended are input into the SVD model, and the user's preference score for the recommended information is obtained; when the user treats the preference score of the recommended information When the recommendation requirement is met, the information to be recommended is recommended to the user. Its reliability is higher, and it can better meet the user's interest requirements.

Description

Information Recommendation Method and Equipment

technical field

The invention relates to the communication field, in particular to an information recommendation method and device.

Background technique

Now is an era of excess information. Every day, users receive a lot of information, but real users like and need very little. Therefore, a technical solution for recommending information to users appears, hoping to recommend the information that the user likes and needs to the user.

At present, there is a scheme for recommending information to users based on content in the current technology. In this prior art, a content-based recommendation system establishes a configuration file for a user. When establishing a user configuration file, it analyzes the content that the user has purchased (or browsed), extracts the historical records and stores them in the user configuration file. When a user purchases or browses new content, update the user profile based on the new content. When recommending content, the system compares the similarity between the user profile and the content to be recommended by calculating the keyword vector of the content and the historical records in the user profile. If there is no matching keyword between the two, it is not New content will be recommended to the user. The content-based content-based recommendation information to the user in the prior art can only use the user's historical records as the recommendation basis, which has poor reliability.

Contents of the invention

Embodiments of the present invention provide an information recommendation method and device, which screen recommended information for users by means of SVD models and social media of users and/or friends, and have high reliability.

The first aspect of the present invention provides an information recommendation method, which may include:

Extract feature factors from the social media of the user and/or the user's friends, which is a network virtual platform for users to compose and share information;

Obtaining the information pointed to by the index information according to the index information in the social media of the user and/or the user's friends;

extracting feature factors from the information pointed to by the index information;

According to the extracted eigenfactors in the social media, the eigenfactors in the information pointed to by the index information, and the user's preference score for the information pointed to by the index information, the singular value SVD model of the user's preference for information is obtained. ;

Input the eigenfactors in the social media and the eigenfactors of the information to be recommended into the SVD model to obtain the user's preference score for the information to be recommended;

When the user's preference score on the information to be recommended meets a recommendation requirement, the information to be recommended is recommended to the user.

With reference to the first aspect, in a first feasible implementation manner, the index information in the social media includes at least one of URLs, titles, and words.

In conjunction with the first aspect, in a second feasible implementation manner, the extraction of feature factors from the social media of the user and/or the user's friends includes: user ID, words in social media, author, article in social media At least one of the number of reposts and the number of comments on articles in the social media.

With reference to the first aspect, in a third feasible implementation manner, the feature factors extracted from the information pointed to by the index information and the feature factors of the information to be recommended include: information source, information content, information type, information At least one of the number of times the information is accessed and the time when the information is accessed.

With reference to any one of the first aspect to the third feasible implementation manner of the first aspect, in a fourth feasible implementation manner, the user's preference score for the information to be recommended meets the recommendation requirements, including:

When the user's preference score for the information to be recommended is greater than a set threshold, it is determined that the recommendation requirement is met;

Alternatively, when the user's preference score for the information to be recommended is the highest relative to the user's preference score for other information to be recommended, it is determined that the recommendation requirement is met.

The second aspect of the embodiment of the present invention provides an information recommendation device, which may include:

The user data extraction module is used to extract feature factors from the social media of the user and/or the user's friends, and the social media is a network virtual platform for users to compose and share information;

An acquisition module, configured to acquire the information pointed to by the index information according to the index information in the social media of the user and/or the user's friends;

An information extraction module, configured to extract feature factors from the information pointed to by the index information;

A model generation module, configured to obtain the user's preference singular value SVD according to the feature factor extracted by the user data extraction module, the feature factor extracted by the information extraction module, and the user's preference score for the information pointed to by the index information Model;

Scoring module, for inputting the eigenfactors in the social media and the eigenfactors of the information to be recommended into the SVD model to obtain the user's preference score for the information to be recommended;

A recommendation module, configured to recommend the information to be recommended to the user when the user's preference score on the information to be recommended meets the recommendation requirements.

With reference to the second aspect, in a first feasible implementation manner, the index information in the social media includes at least one of URLs, titles, and words.

In conjunction with the second aspect, in a second feasible implementation manner, the extraction of feature factors from the social media of the user and/or the user's friends includes: user ID, words in social media, authors, articles in social media At least one of the number of reposts and the number of comments on articles in the social media.

With reference to the second aspect, in a third feasible implementation manner, the feature factors extracted from the information pointed to by the index information and the feature factors of the information to be recommended include: information source, information content, information type, information At least one of the number of times the information is accessed and the time when the information is accessed.

With reference to any one of the second aspect to the third feasible implementation manner of the second aspect, in a fourth feasible implementation manner, the user's preference score for the information to be recommended meets the recommendation requirements, including:

When the user's preference score for the information to be recommended is greater than a set threshold, it is determined that the recommendation requirement is met;

Alternatively, when the user's preference score for the information to be recommended is the highest relative to the user's preference score for other information to be recommended, it is determined that the recommendation requirement is met.

As can be seen from the above, in some feasible embodiments of the present invention, feature factors are extracted from the social media of the user and/or the user's friends, and the social media is a network virtual platform for users to compose and share information; according to the user And/or the index information in the social media of the user's friend acquires the information pointed to by the index information; extracts the feature factor from the information pointed to by the index information; according to the extracted feature factor in the social media and The feature factors in the information pointed to by the index information and the user's preference score for the information pointed to by the index information are obtained to obtain the user's preference singular value SVD model; the feature factors in the social media and the information to be recommended The eigenfactors are input into the SVD model to obtain the user's preference score for the information to be recommended; when the user's preference score for the information to be recommended meets the recommendation requirements, recommend the user to recommend the information to be recommended information. Its reliability is higher, and it can better meet the interest requirements of users.

Description of drawings

FIG. 1 is a schematic flow diagram of an embodiment of the information recommendation method of the present invention;

FIG. 2 is a schematic diagram of the structural composition of an embodiment of the information recommendation device of the present invention;

FIG. 3 is a schematic diagram of the structural composition of another embodiment of the information recommendation device of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings.

FIG. 1 is a schematic flowchart of an embodiment of the information recommendation method of the present invention. As shown in Figure 1, the method of the present invention may include:

Step S110, extracting feature factors from the social media of the user and/or the user's friends, the social media is a network virtual platform for users to compose and share information.

In specific implementation, the social media in the embodiment of the present invention may include: Weibo, WeChat, Facebook and other network virtual platforms for users to compose and share information.

In a specific implementation, the feature factor extracted from the social media of the user and/or the user's friend in the embodiment of the present invention can be any information that can be obtained from the social media. As a feasible implementation mode, the user and/or the user's friend The feature factors extracted from social media can include user ID (which can be used to uniquely identify a user), words in social media (for example, words that appear more than a predetermined number of times, current popular Internet terms, etc.), author , at least one of the number of reposts of articles in social media, the number of comments of articles in social media, and the like.

Step S111, according to the index information in the social media of the user and/or the user's friend, obtain the information pointed to by the index information.

In a specific implementation, the index information in social media may include URLs, titles, words (for example, words whose frequency of occurrence exceeds a predetermined number of times, current popular Internet terms, etc.) and the like. Then in step S111, the content pointed to by the index information can be further obtained according to these index information, for example, the news webpage linked by the URL can be opened, the microblog article of the friend linked by the URL can be opened; Enter the specific article of the title; as another example, you can use the word as an index to search for more information related to the word.

Step S112, extracting feature factors from the information pointed to by the index information.

In a specific implementation, the feature factor extracted from the information pointed to by the index information can be any information that can be obtained from the information. As a feasible implementation, the feature factor extracted from the information pointed to by the index information The feature factors may include: at least one of information source, information content (for example, words that appear more than a predetermined number of times, current popular Internet terms, etc.), information type, information access times, and information access time. Taking the news opened by the website as an example, the information source of the news can be extracted in step S112, such as the news released by Xinhua News Agency, and the information content in the news is extracted, such as GDP, house price regulation; the type of news extracted: for example, finance, Economy; characteristic factors such as the number of news visits, for example, 3000 times.

Step S113, according to the extracted feature factors in the social media, the feature factors in the information pointed to by the index information, and the user's preference score for the information pointed to by the index information to obtain the user's preference for information Value (Singular Value Decomposition, SVD) model.

Step S114, input the feature factors in the social media and the feature factors of the information to be recommended into the SVD model to obtain the user's preference score for the information to be recommended.

In a specific implementation, the feature factor extracted from the information to be recommended may be consistent with the feature factor extracted from the information pointed to by the index information, such as including: information source, information content (for example, the frequency of occurrence exceeds a predetermined number of times At least one of the following words, current popular Internet terms, etc.), type of information, number of times information is accessed, and time when information is accessed.

Step S115, when the user's preference score on the information to be recommended meets the recommendation requirement, recommend the information to be recommended to the user.

In a specific implementation, when the SVD model is generated, it is assumed that the user is interested in all the information pointed to by the index information, so these information are given preference scores that meet the recommendation requirements.

In the specific implementation, there are many ways to realize the SVD algorithm. The embodiment of the present invention utilizes the feature factors in the social media and the feature factors in the information pointed to by the index information in the social media and the user's prior reference to the index information pointed to by the user. The information preference score solves the SVD model of the user's preference for information, and thus evaluates the user's preference for the information in the entire network through the user's preference score for the existing information. Compared with the existing technology, the user history record is the key The method of word matching can obtain more accurate results of user preferences, thereby recommending information to users, which has higher reliability and can better meet the user's interest requirements.

In a specific implementation, whether the recommendation requirements are met can be set arbitrarily in advance according to the requirements. For example, it can be set: when the user's preference score for the information to be recommended is greater than the set threshold, it is determined that the recommendation requirements are met; or, when When the user's preference score for the information to be recommended is the highest relative to the user's preference score for other information to be recommended, it is determined that the recommendation requirement is met.

FIG. 2 is a schematic structural diagram of an embodiment of an information recommendation device of the present invention. As shown in Figure 2, it may include a user data extraction module 21, an acquisition module 22, an information extraction module 23, a model generation module 24, a scoring module 25 and a recommendation module 26, wherein:

The user data extraction module 21 is configured to extract feature factors from the social media of the user and/or the user's friends, and the social media is a network virtual platform for users to compose and share information.

In specific implementation, the social media in the embodiment of the present invention may include: Weibo, WeChat, Facebook and other network virtual platforms for users to compose and share information.

In a specific implementation, the user data extraction module 21 in the embodiment of the present invention extracts the feature factor from the social media of the user and/or the user's friends, which can be any information that can be obtained from the social media. As a feasible implementation, the The feature factors extracted from the social media of the user and/or the user's friends may include user ID (which can be used to uniquely identify a user), words in social media (for example, words with a frequency of occurrence exceeding a predetermined number of times, currently popular Internet terms, etc.), author, number of reposts of articles in social media, number of comments of articles in social media, and the like.

The obtaining module 22 is configured to obtain the information pointed to by the index information according to the index information in the social media of the user and/or the user's friends.

In a specific implementation, the index information in social media may include URLs, titles, words (for example, words whose frequency of occurrence exceeds a predetermined number of times, current popular Internet terms, etc.) and the like. Then the acquisition module 22 can further acquire the content pointed to by the index information according to these index information, for example, the news webpage linked by the URL can be opened, the microblog article of the friend linked by the URL can be opened; The specific article of the title; as another example, the word can be used as an index to search for more information related to the word.

An information extraction module 23, configured to extract feature factors from the information acquired by the acquisition module.

In a specific implementation, the feature factor extracted by the information extraction module 23 from the information pointed to by the index information can be any information that can be obtained from the information. As a feasible implementation mode, from the information pointed to by the index information The feature factors extracted from the information may include: information source, information content (for example, words that appear more than a predetermined number of times, current popular Internet terms, etc.), information type, information access times, information access time at least A sort of. Taking the news opened by the website as an example, the information source of the news can be extracted in step S112, such as the news released by Xinhua News Agency, and the information content in the news is extracted, such as GDP, house price regulation; the type of news extracted: for example, finance, Economy; characteristic factors such as the number of news visits, for example, 3000 times.

A model generation module 24, configured to obtain user preferences according to the feature factors extracted by the user data extraction module 21, the feature factors extracted by the information extraction module 23, and the user's preference score for the information pointed to by the index information Singular value SVD model.

In a specific implementation, when the SVD model is generated, it is assumed that the user is interested in all the information pointed to by the index information, so these information are given preference scores that meet the recommendation requirements.

The scoring module 25 is configured to input the eigenfactors extracted by the user data extraction module 21 and the eigenfactors of the information to be recommended into the SVD model to obtain the user's preference score for the information to be recommended.

The recommendation module 26 is configured to recommend the information to be recommended to the user when the user's preference score on the information to be recommended meets the recommendation requirements.

In the specific implementation, there are many ways to realize the SVD algorithm. The embodiment of the present invention utilizes the feature factors in the social media and the feature factors in the information pointed to by the index information in the social media and the user's prior reference to the index information pointed to by the user. The information preference score solves the SVD model of the user's preference for information, and thus evaluates the user's preference for the information in the entire network through the user's preference score for the existing information. Compared with the existing technology, the user history record is the key The method of word matching can obtain more accurate results of user preferences, thereby recommending information to users, which has higher reliability and can better meet the user's interest requirements.

In a specific implementation, whether the recommendation requirements are met can be set arbitrarily in advance according to the requirements. For example, it can be set: when the user's preference score for the information to be recommended is greater than the set threshold, it is determined that the recommendation requirements are met; or, when When the user's preference score for the information to be recommended is the highest relative to the user's preference score for other information to be recommended, it is determined that the recommendation requirement is met.

Fig. 3 is a schematic structural diagram of another embodiment of the information recommendation device of the present invention. The difference between this embodiment and the embodiment described in FIG. 2 is that what is shown is the composition structure of the hardware modules of the device. As shown in FIG. 3 , the information recommendation device of the present invention may include a memory 31, a processor 32, and a display 33 in terms of hardware composition, wherein: the memory 31 stores a program, and the processor 32 is used to call the memory The program stored in 31 is also used to execute the functions realized by the user data extraction module 21, acquisition module 22, information extraction module 23, model generation module 24 and scoring module 25 shown in Figure 2, and the display 33 is used when the When the user's preference score on the information to be recommended meets the recommendation requirement, the information to be recommended is recommended to the user.

The above-listed are only preferred embodiments of the present invention, which of course cannot limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention still fall within the scope of the present invention.

Claims (10)

1. an information recommendation method, is characterized in that, comprising:

From the social media of user and/or user good friend, extract characterization factor, described social media is write the net virtual platform with sharing information for user;

The information pointed by described index information is obtained according to the index information in the social media of described user and/or described user good friend;

Characterization factor is extracted from the information pointed by described index information;

Characterization factor in information pointed by the characterization factor of the extraction in described social media and described index information and the hobby scoring of described user to the information pointed by described index information obtain the hobby singular value SVD model of user to information;

The characterization factor of the characterization factor in described social media and information to be recommended is inputted described SVD model, obtains described user and the hobby of described information to be recommended is marked;

When described user to the scoring of the hobby of described information to be recommended meet recommendation require time, recommend described information to be recommended to described user.

2. information recommendation method as claimed in claim 1, it is characterized in that, the index information in described social media comprises at least one in network address, title, words.

3. information recommendation method as claimed in claim 1, it is characterized in that, describedly from the social media of user and/or user good friend, extract characterization factor comprise: at least one in the comment number of the article in the forwarding number of the article in the words in user ID, social media, author, social media, described social media.

4. information recommendation method as claimed in claim 1, it is characterized in that, the described characterization factor extracting characterization factor and information to be recommended from the information pointed by described index information comprises: the accessed number of times of the type of information source, the information content, information, information, at least one in the information accessed time.

5. the information recommendation method according to any one of claim 1-4, is characterized in that, the hobby scoring of described user to described information to be recommended meets recommends requirement, comprising:

When the hobby scoring of described user to described information to be recommended is greater than the threshold value of setting, determines to meet and recommend requirement;

Or, when the hobby of user to described information to be recommended marks the hobby scoring of relative user to other information to be recommended for time the highest, determine to meet and recommend requirement.

6. an information recommendation equipment, is characterized in that, comprising:

User data extraction module, for extracting characterization factor in the social media from user and/or user good friend, described social media is write the net virtual platform with sharing information for user;

Acquisition module, for obtaining the information pointed by described index information according to the index information in the social media of described user and/or described user good friend;

Information extraction modules, for extracting characterization factor from the information pointed by described index information;

Model generation module, the characterization factor extracted for the characterization factor that extracts according to described user data extraction module and described information extraction modules and the hobby scoring of described user to the information pointed by described index information obtain the hobby singular value SVD model of user;

Grading module, for the characterization factor of the characterization factor in described social media and information to be recommended is inputted described SVD model, obtains described user and marks to the hobby of described information to be recommended;

Recommending module, for when described user to the scoring of the hobby of described information to be recommended meet recommendation require time, recommend described information to be recommended to described user.

7. information recommendation equipment as claimed in claim 6, it is characterized in that, the index information in described social media comprises at least one in network address, title, words.

8. information recommendation equipment as claimed in claim 6, it is characterized in that, describedly from the social media of user and/or user good friend, extract characterization factor comprise: at least one in the comment number of the article in the forwarding number of the article in the word in user ID, social media, author, social media, described social media.

9. information recommendation equipment as claimed in claim 6, it is characterized in that, the described characterization factor extracting characterization factor and information to be recommended from the information pointed by described index information comprises: the accessed number of times of the type of information source, the information content, information, information, at least one in the information accessed time.

10. the information recommendation equipment according to any one of claim 6-9, is characterized in that, the hobby scoring of described user to described information to be recommended meets recommends requirement, comprising:

When the hobby scoring of described user to described information to be recommended is greater than the threshold value of setting, determines to meet and recommend requirement;

Or, when the hobby of user to described information to be recommended marks the hobby scoring of relative user to other information to be recommended for time the highest, determine to meet and recommend requirement.